Disable ocr in tests previously marked as integration

tests/cassettes/test_converters/TestDoclingLocalConverter.test_picture_description_end_to_end.yaml

@@ -28,7 +28,7 @@ interactions:
   response:
     headers:
       content-length:
-      - '1674'
+      - '1931'
       content-type:
       - application/json
     parsed_body:
@@ -37,48 +37,37 @@ interactions:
         index: 0
         message:
           content: |-
-            This is a screenshot of a software interface for labeling or tagging elements within a document.
+            This is a screenshot of a software interface used for labeling or annotating elements within a document or image.
 
-            Here’s a concise description for better understanding:
+            The screenshot depicts a tool designated for identifying and classifying document elements using field labels listed on the right side. Here's a concise breakdown of the main elements:
 
-            1. **Main Area**:
+            1. **Field Labels:**
-               - There is a preview space displaying an arrangement of colored blocks and shapes, likely representing parts of a document layout.
+               - The side panel (at the right) displays a list of category options for types of document elements. These include: Text, Picture, Formula, Code, Complex-form (highlighted), Section Header, Page Header, Page Footer, Footnote, Table, List Item, and Title. One category, "Complex-form," is currently highlighted.
-               - The colored areas include horizontal bars (various colors: green, red, blue, yellow, pink), graphical charts, and some abstract text or layout boxes. There is also a pie chart and a pair of bar-like shapes.
 
-            2. **Field Labels Box**:
+            2. **Document Display Area:**
-               A section on the right side labeled "Field labels."
+               - In the center and on the left side of the image, there are various colored sections resembling different parts of a document format like spreadsheets, charts, forms, graphics, and a color-coded bar-like structure.
-               The instructions read: “Identify the document elements using the buttons below.”
+               - Each section has a cursor or blue anchor point (with an "X"), which you likely use to select and annotate an element.
+               - Specific color-coded elements include:
+                 - Horizontal and vertical color bars
+                 - Tables and form grid
+                 - Pie chart and bar chart graphics
+                 - Sections of a form (in red and grey)
 
-            3. **Label Options**:
+            3. **Tool Buttons/Bars:**
-               - A list of field labels is given as a set of radio button selections, including:
+               - At the bottom left, there are interface buttons labeled "Skip", "Filter", and "Submit".
-                 - Text
+               - "Report Problem" is a button found at the bottom right of the interface.
-                 - Picture
-                 - Formula
-                 - Code
-                 - Complex-form
-                 - Section-header
-                 - Page-footer
-                 - Page-header
-                 - Footnote
-                 - Table
-                 - List-item
-                 - Title
 
-            4. **Navigation Elements**:
+            The visible text primarily appears in this list of category options, highlighted labels, and button texts. The specific content within the graphic segments is not readable text but visual representations.
-               There are three buttons at the bottom left: "Skip," "Filter," and "Submit."
-               There is additionally a highlighted "Report problem" button at the bottom right, in blue.
-
-            The elements depicted visually serve to assist in identifying and tagging different parts of the document interface for further processing or analysis.
           role: assistant
-      created: 1767616488
+      created: 1768227488
-      id: chatcmpl-151
+      id: chatcmpl-657
       model: ministral-3
       object: chat.completion
       system_fingerprint: fp_ollama
       usage:
-        completion_tokens: 290
+        completion_tokens: 339
         prompt_tokens: 836
-        total_tokens: 1126
+        total_tokens: 1175
     status:
       code: 200
       message: OK

tests/test_client.py

@@ -1098,11 +1098,16 @@ async def test_client_visualize_chunk_with_pdf(temp_db_path):
     """Test visualize_chunk returns images with bounding boxes for PDF documents."""
     from PIL.Image import Image as PILImage
 
+    from haiku.rag.config import AppConfig
+
     pdf_path = Path("tests/data/doclaynet.pdf")
     if not pdf_path.exists():
         pytest.skip("doclaynet.pdf not found")
 
-    async with HaikuRAG(temp_db_path, create=True) as client:
+    config = AppConfig()
+    config.processing.conversion_options.do_ocr = False
+
+    async with HaikuRAG(temp_db_path, config=config, create=True) as client:
         doc = await client.create_document_from_source(pdf_path)
         assert isinstance(doc, Document)
         assert doc.id is not None

tests/test_context_enhancement.py

@@ -761,6 +761,7 @@ async def test_expand_context_no_base64_images_docling_local(temp_db_path):
     config = AppConfig()
     config.processing.converter = "docling-local"
     config.processing.chunker = "docling-local"
+    config.processing.conversion_options.do_ocr = False
     config.search.context_radius = 5
 
     async with HaikuRAG(temp_db_path, config=config, create=True) as client:

tests/test_converters.py

@@ -407,6 +407,8 @@ class TestDoclingLocalConverter:
         if not pdf_path.exists():
             pytest.skip("doclaynet.pdf not found")
 
+        # Disable OCR (not needed for native PDF, avoids model downloads)
+        config.processing.conversion_options.do_ocr = False
         # Enable picture description with Ollama
         config.processing.conversion_options.picture_description.enabled = True
         config.processing.conversion_options.picture_description.model.provider = (